A geostatistical approach to down-scale climate forecasts

Posted 17 February 2013

A storm from space (credit: NASA).

Researchers have found a new technique for generating small-scale climate models from large-scale climate models.

While general models of global circulation are the tool of choice for forecasting the effects of climate change, their spatial resolutions are too broad for the needs of regional planners.

To provide locally-relevant information, modellers typically employ one of two techniques: producing a new forecast using a regional dynamic model, or statistically down-scaling the projections of the larger model.

Jha et al. have proposed a geostatistical technique to translate climate modelling results onto a smaller spatial scale.

This approach geostatistical approach used climate model output from a regional dynamical model for the years 1985 to 2005 to create an image of the Murray-Darling Basin, a 1 million square kilometre watershed in south eastern Australia.

From the model data, the authors identified spatial patterns for three variables: the latent heat flux, soil moisture, and the temperature of the uppermost land surface layer.

The regional model used could produce projections at two different resolutions, with either 50 or 10 square kilometre grid cells.

For the years 1985 to 2005, the authors used model calculations at both resolutions to train their model and learn the relationship between large- and small-scale patterns. For 2006, the authors used their geostatistical technique to statistically downscale the 50 kilometre-resolution model output onto a 10 kilometre grid. They used the 10 kilometre-resolution model output to check their approach.

The authors found that for latent heat flux, surface temperature, and soil moisture, their approach did a good job down-scaling the 50-kilometre model output.

For soil moisture, high resolution artefacts such as rivers and lakes unavoidably impacted the downscaled response.

The authors suggest that although they used a high-resolution regional model to determine spatial patterns, the same could be done using remote sensing techniques.

The technique also has the potential in feature sharpening in remote sensing applications through image fusion, filling gaps in spatial data, and aggregating/disaggregating hydrological and groundwater variables for catchment studies.

Links

Latest news

Groundwater resources in Africa resilient to climate change

Groundwater resources in Africa resilient to climate change

8 August 2019

Groundwater – a vital source of water for drinking and irrigation across sub-Saharan Africa – is resilient to climate variability and change, according to a new study.

Read more…

 Low cost way to explore groundwater resources could be game changer

Low cost way to explore groundwater resources could be game changer

10 May 2019

UNSW Sydney water engineers have revealed that investigating and managing groundwater resources more sustainably can be achieved at lower cost by using existing Earth and atmospheric tidal data.

Read more…

CWI Director addressed National Ground Water Monitoring issues in ABC News

CWI Director addressed National Ground Water Monitoring issues in ABC News

5 March 2019

The Director of CWI was recently quoted on the ABC story: ‘ Who's watching the water? Experts sound warning on deteriorating groundwater monitoring’

Read more…

Spotlight on mining and water with Dr Wendy Timms

Spotlight on mining and water with Dr Wendy Timms

10 November 2017

Dr Wendy Timms talks about her many hats and why diversity is critical in sustainable mining practice.

Read more…

Groundwater ‘pit stops’ enabled survival and migration of our ancient ancestors

Groundwater ‘pit stops’ enabled survival and migration of our ancient ancestors

1 June 2017

African groundwater helped kick-start the evolutionary history of humans, with the movement of our ancestors across East Africa shaped by the location of springs, new research suggests.

Read more…